[Proposal] Team Hana Gabrielle — Fix p5.js benchmark cleanup

Submissions_SOSE_2026/proposal-team-hanagabrielle.md

@@ -0,0 +1,90 @@
+# Proposal Submission — SOSE 2026
+
+---
+
+## Team Details* 👥
+
+**Mentor's Name:** Hana Gabrielle Bidon
+
+**Team Members:**
+
+| GitHub Username | Full Name            |
+|-----------------|----------------------|
+| @Danyccsf       | Dany Valverde Caldas |
+
+---
+
+## p5.js Issue* 📋
+
+- **Link to the issue:** https://github.com/processing/p5.js/issues/9041
+
+- Issue Title: Benchmark cleanup passes `p5.model` to `freeGeometry`
+
+- Repository: p5.js
+
+---
+
+## Abstract* 📝
+
+My proposal fixes a reproducible benchmark-cleanup defect in p5.js. During local testing, I found that `freeGeometry()` received the `myp5.model` function reference instead of the geometry instance created by `buildGeometry()`, emitting a Friendly Error System (FES) validation message while the benchmark continued. The scoped change passes the generated `shape` instance to `freeGeometry()` and is verified through the affected benchmark. This work corrects invalid cleanup usage without changing p5.js or FES behavior more broadly.
+
+---
+
+## Problem Statement* 📑
+
+In `test/bench/cpu_transforms.bench.js`, the cleanup call passed `myp5.model`—a function reference—to `freeGeometry()` rather than the `shape` geometry instance. This invalid API usage emitted an FES validation message while the benchmark continued to report results. Passing `shape` instead removes the observed message in isolated WebGL and WebGPU benchmark runs. Contributors and maintainers need the benchmark to execute its intended cleanup call correctly while evaluating CPU-transform behavior.
+
+---
+
+## Proposed Solution* 💡
+
+I will submit the verified one-line correction in `test/bench/cpu_transforms.bench.js`:
+
+```js
+myp5.freeGeometry(shape);
+```
+
+I will run the targeted CPU-transforms benchmark in its WebGL and WebGPU configurations before submitting the pull request. This narrow approach directly corrects the reported defect and follows the maintainer’s request to create Issue #9041 before opening a PR.
+
+---
+
+## Research on old issues and Maintainer Patterns* 🔭
+
+Issue https://github.com/processing/p5.js/issues/8953 documents ongoing work on p5.strands transformation support across graphics backends. It is relevant background for the project’s graphics direction, but this proposal does not implement its assigned transformation-matrix work.
+
+I also reviewed https://github.com/processing/p5.js/pull/8887, which modularized the Friendly Error System. My benchmark finding shows that invalid API usage can emit an FES validation message while a benchmark still passes. I will not assume that all FES messages should globally fail benchmarks; this proposal corrects the specific misuse tracked in https://github.com/processing/p5.js/issues/9041.
+
+---
+
+## Impact* 🛠️
+
+- [X] **Bug Fix** — Corrects a verified invalid `freeGeometry()` call in `cpu_transforms.bench.js`.
+- [X] **Testing** — Verifies the correction through the affected WebGL and WebGPU benchmark runs.
+
+---
+
+## Inclusivity and Accessibility 🤝
+
+Not applicable — bug fix.
+
+---
+
+## Implementation Plan* ⏳
+
+- **Week 5 (Completed)**: Reproduce the benchmark message, identify the incorrect argument, and prepare the one-line correction.
+- **Week 6 (Completed)**: Discuss the finding with @davepagurek and create Issue #9041 as requested.
+- **Week 7**: Create the focused pull request, run the targeted WebGL and WebGPU benchmark validation, and request review.
+- **Week 8**: Respond to review feedback and document the result in the final technical presentation.
+
+---
+
+## Deliverables* 📦
+
+- A pull request correcting the cleanup call in `test/bench/cpu_transforms.bench.js`.
+- Targeted benchmark results for the WebGL and WebGPU configurations showing that the prior FES message no longer appears.
+
+---
+
+## Anything Else?
+
+I discussed the finding with @davepagurek in the p5.js Discord before opening Issue #9041. The implementation is intentionally limited to the maintainer-confirmed correction.